Evolutionary-based Image Segmentation Methods

Nearly 72 years ago, Wertheimer [1] pointed out the importance of perceptual grouping and organization in vision and listed several key factors, such as similarity, proximity, and good continuation, which lead to visual grouping. However, even to this day, many of the computational issues of perceptual grouping have remained unresolved. Since there are many possible partitions of an image into subsets, how do you know which one is right? There are two aspects to be considered here. The first is that there may not be a single correct answer. The second aspect is that the partitioning is inherently hierarchical. Prior literature on the related problems of clustering, grouping and image segmentation is huge. Unfortunately, there is not a general method existing to solve the problem.[2] Image segmentation is one of the central problems in computer vision and pattern recognition. It refers to the process of assigning a label to every pixel in an image such that pixels with the same label share certain visual characteristics. The result of image segmentation is a set of segments (sets of pixels) that collectively cover the entire image. Pixels in the same region are similar with respect to some characteristics or computed properties, such as color, intensity, and texture. Adjacent regions are significantly different with respect to the same characteristics. The goal of segmentation is to simplify and/or change the representation of an image into something that is more meaningful and easier to analyze.[3] There are many general-purpose approaches available for image segmentation such as threshold methods[4], edge-based methods[5], region-based methods[6], and graph-based methods[7]. Threshold techniques make decisions based on local pixel information. Edgebased methods are based on connecting together broken contour lines. It is prone to failure in the presence of blurring. A region-based method usually partitions an image into connected regions by grouping neighboring pixels of similar intensity levels. Adjacent regions are then merged under some characteristics. Graph-based techniques generally represent the problem in terms of a graph where each node corresponds to a pixel in the image, and an edge connects each pair of vertices. A weight is associated with each edge based on some property of the pixels that it connects, such as their image intensities. Hybrid techniques using a mix of the methods above are also popular. What listed above also exposed two basic questions: • What is the precise criterion for a good segmentation? • How can such a segmentation be computed efficiently?